Abstract

Image processing and image analysis tasks have large data processing requirements and inherent parallelism and are well suited to implementation on digital optical processors because of the parallelism and free interconnection capabilities of optical systems [1][2]. Recently, several techniques for constructing optical cellular logic processors for image processing have been proposed [2]-[5]. Through parallel studies of architectures, algorithms, mathematical structures, and optics we have found that: 1) cellular automata are appropriate models for parallel image processing machines [6]; 2) an image algebra extending from mathematical morphology [7] [8] can lead to a formal parallel language approach to the design of image processing algorithms; 3) the algebraic structure serves as a framework for both algorithms and architectures of parallel image processing; and 4) optical computing techniques are able to efficiently implement image algebra based on cellular logic architectures (e.g. cellular array, cellular hypercube etc.). Here we will first discuss image algebra and then architectures for its implementation.

© 1987 Optical Society of America